Remote Settings
The remote-settings.sys.mjs module offers the ability to fetch remote settings that are kept in sync with Mozilla servers.
Usage
The get()
method returns the list of entries for a specific key. Each entry can have arbitrary attributes, and can only be modified on the server.
const { RemoteSettings } = ChromeUtils.importESModule("resource://services-settings/remote-settings.sys.mjs");
const data = await RemoteSettings("a-key").get();
/*
data == [
{label: "Yahoo", enabled: true, weight: 10, id: "d0782d8d", last_modified: 1522764475905},
{label: "Google", enabled: true, weight: 20, id: "8883955f", last_modified: 1521539068414},
{label: "Ecosia", enabled: false, weight: 5, id: "337c865d", last_modified: 1520527480321},
]
*/
for (const entry of data) {
// Do something with entry...
// await InternalAPI.load(entry.id, entry.label, entry.weight);
});
Note
The id
and last_modified
(timestamp) attributes are assigned by the server.
Empty local database
On new user profiles or for recently added use-cases, the local database will be empty until a synchronization with the server happens. Synchronizations are managed internally, and can sometimes be triggered minutes after browser starts.
By default, if .get()
is called before the local database had the chance to be synchronized, and if no initial data was provided (see below), then the settings will be pulled from the server in order to avoid returning an empty list. In that case, the first call to .get()
will thus take longer than the following ones.
This behaviour can be disabled using the syncIfEmpty
option.
Important
If the implicit synchronization fails (e.g network is not available) then errors are silent and an empty list is returned. Uptake Telemetry status is sent though.
Options
filters
,order
: The list can optionally be filtered or ordered:const subset = await RemoteSettings("a-key").get({ filters: { property: "value" }, order: "-weight" });
syncIfEmpty
: iftrue
and no local data is present, then look for a packaged dump to load. If none, then pull records from network. Set it tofalse
to skip loading of packaged dump and network activity. Use this only if your use-case can tolerate an empty list until the first synchronization happens (default:true
).await RemoteSettings("a-key").get({ syncIfEmpty: false });
verifySignature
: verify the content signature of the local data (default:false
). An error is thrown if the local data was altered. This hurts performance, but can be used if your use case needs to be secure from local tampering.emptyListFallback
: return an empty list if obtaining the records fails (default:true
). If an error occurs during the reading of local data, or during synchronization, then an empty list is returned.
Events
The on()
function registers handlers to be triggered when events occur.
The "sync"
event allows to be notified when the remote settings are changed on the server side. Your handler is given an event
object that contains a data
attribute that has information about the changes:
current
: current list of entries (after changes were applied);created
,updated
,deleted
: list of entries that were created/updated/deleted respectively.
RemoteSettings("a-key").on("sync", event => {
const { data: { current } } = event;
for (const entry of current) {
// Do something with entry...
// await InternalAPI.reload(entry.id, entry.label, entry.weight);
}
});
Note
Currently, the synchronization of remote settings is triggered via push notifications, and also by its own timer every 24H (see the preference services.settings.poll_interval
).
File attachments
When an entry has a file attached to it, it has an attachment
attribute, which contains the file related information (url, hash, size, mimetype, etc.).
Remote files are not downloaded automatically. In order to keep attachments in sync, the provided helper can be leveraged like this:
const client = RemoteSettings("a-key");
client.on("sync", async ({ data: { created, updated, deleted } }) => {
const toDelete = deleted.filter(d => d.attachment);
const toDownload = created
.concat(updated.map(u => u.new))
.filter(d => d.attachment);
// Remove local files of deleted records
await Promise.all(
toDelete.map(record => client.attachments.deleteDownloaded(record))
);
// Download a bundle of all attachments if local cache is empty (see details below)
client.attachments.cacheAll();
// OR download new attachments individually
const fileContents = await Promise.all(
toDownload.map(async record => {
const { buffer } = await client.attachments.download(record);
return buffer;
});
);
});
- The provided
cacheAll
helper will: be a no-op if the local attachment cache is not empty - Use
pruneAttachments()
to clear or set theforce
parameter to true if you know you want to override this.download and extract all attachments if bundling is enabled for the collection
return a nullable boolean to inform the calling function what happened -
null
if the attachment bundle download was not attempted (ex: client is offline) -false
if at least one attachment failed to be extracted -true
if the bundle was found and extracted without error
- The provided
download
helper will: fetch the remote binary content
write the file in the local IndexedDB
check the file size
check the content SHA256 hash
do nothing if the attachment was already present and sound locally.
Important
The following aspects are not taken care of (yet! help welcome):
check available disk space
preserve bandwidth
resume downloads of large files
Note
The download()
method supports the following options:
retries
(default:3
): number of retries on network errorsfallbackToCache
(default:false
): allows callers to fall back to the cached file and record, if the requested record’s attachment fails to download. This enables callers to always have a valid pair of attachment and record, provided that the attachment has been retrieved at least once.fallbackToDump
(default:false
): activates a fallback to a dump that has been packaged with the client, when other ways to load the attachment have failed. See services/packaging-attachments for more information.
Note
A downloadAsBytes()
method returning an ArrayBuffer
is also available, if writing the attachment locally is not necessary.
Some downloadToDisk()
and deleteFromDisk()
methods are also available but generally discouraged, since they are prone to leaving extraneous files
in the profile directory (see Bug 1634127).
Initial data
It is possible to package a dump of the server records that will be loaded into the local database when no synchronization has happened yet.
The JSON dump will serve as the default dataset for .get()
, instead of doing a round-trip to pull the latest data. It will also reduce the amount of data to be downloaded on the first synchronization.
Place the JSON dump of the server records in the
services/settings/dumps/main/
folder. Using this command:CID="your-collection" curl "https://firefox.settings.services.mozilla.com/v1/buckets/main/collections/${CID}/changeset?_expected=0" | jq '{"data": .changes, "timestamp": .timestamp}' > services/settings/dumps/main/${CID}.json``
Add the filename to the relevant
FINAL_TARGET_FILES
list inservices/settings/dumps/main/moz.build
Please consider the application(s) where the collection is used and only include the dump file in the relevant builds.
If it is only for Firefox desktop, i.e.
browser/
, then add it to a build-specific browser section.If it is for all applications, i.e. outside of
browser/
are other specific area, then add it to the global section.
Now, when RemoteSettings("some-key").get()
is called from an empty profile, the some-key.json
file is going to be loaded before the results are returned.
JSON dumps in the tree are periodically updated by taskcluster/docker/periodic-updates/scripts/periodic_file_updates.sh
.
If your collection’s in-tree dump should not be kept up to date by this automation, place the JSON file in services/settings/static-dumps/
instead.
Note
The example above uses “main” because that’s the default bucket name. If you have customized the bucket name, use the actual bucket name instead of “main”.
Packaging attachments
Attachments are not included in the JSON dumps by default. You may choose to package the attachment with the client, for example if it is important to have the data available at the first startup without requiring network activity. Or if most users would download the attachment anyway. Only package attachments if needed, since they increase the file size of the Firefox installer.
To package an attachment for consumers of the download() method:
Select the desired attachment record from the JSON dump of the server records, and place it at
services/settings/dumps/<bucket name>/<collection name>/<attachment id>.meta.json
. The<attachment id>
defaults to theid
field of the record. If thisid
is not fixed, you must choose a custom ID that can be relied upon as a long-term attachment identifier. See the notes below for more details.Download the attachment associated with the record, and place it at
services/settings/dumps/<bucket name>/<collection name>/<attachment id>
.Update
taskcluster/docker/periodic-updates/scripts/periodic_file_updates.sh
and add the attachment, by editing thecompare_remote_settings_files
function and describing the attachment. Unlike JSON dumps, attachments must explicitly be listed in that update script, because the attachment selection logic needs to be codified in ajq
filter in the script. For an example, see Bug 1636158.Register the location of the
<attachment id>.meta.json
and<attachment id>
in themoz.build
file of the collection folder, and possiblypackage-manifest.in
, as described in the previous section about registering JSON dumps <services/initial-data>.
Note
<attachment id>
is used to derive the file names of the packaged attachment dump, and as the
key for the cache where attachment updates from the network are saved.
The attachment identifier is expected to be fixed across client application updates.
If that expectation cannot be met, the attachmentId
option of the download
method of the
attachment downloader should be used to override the attachment ID with a custom (stable) value.
In order to keep track of the cached attachment, and prevent it from being pruned automatically,
the attachment identifier will have to be explicitly listed in the keepAttachmentsIds = [<attachment id>]
option of the RemoteSettings client constructor.
Note
The contents of the .meta.json
file is already contained within the records, but separated
from the main set of records to ensure the availability of the original record with the data,
independently of the packaged or downloaded records.
This file may become optional in a future update, see Bug 1640059.
Synchronization Process
The synchronization process consists in pulling the recent changes, merging them with the local data, and verifying the integrity of the result.
Important
As shown above, we can end-up in situations where synchronization fails and will leave the local DB in an empty state.
Targets and A/B testing
In order to deliver settings to subsets of the population, you can set targets on entries (platform, language, channel, version range, preferences values, samples, etc.) when editing records on the server.
From the client API standpoint, this is completely transparent: the .get()
method — as well as the event data — will always filter the entries on which the target matches.
Note
The remote settings targets follow the same approach as the Normandy recipe client (ie. JEXL filter expressions).
Uptake Telemetry
Some uptake telemetry is collected in order to monitor how remote settings are propagated.
It is submitted to a single keyed histogram whose id is UPTAKE_REMOTE_CONTENT_RESULT_1
and the keys are prefixed with main/
(eg. main/a-key
in the above example).
Create new remote settings
Staff members can create new kinds of remote settings, following this documentation.
It basically consists in:
Choosing a key (eg.
search-providers
)Assigning collaborators to editors and reviewers groups
(optional) Define a JSONSchema to validate entries
(optional) Allow attachments on entries
And once done:
Create, modify or delete entries and let reviewers approve the changes
Wait for Firefox to pick-up the changes for your settings key
Global Notifications
The polling for changes process sends two notifications that observers can register to:
remote-settings:changes-poll-start
: Polling for changes is starting. triggered either by the scheduled timer or a push broadcast.remote-settings:changes-poll-end
: Polling for changes has endedremote-settings:sync-error
: A synchronization error occurred. Notification subject provides information about the error and affected collection in thewrappedJSObject
attribute.remote-settings:broken-sync-error
: Synchronization seems to be consistently failing. Profile is at risk.
const observer = {
observe(aSubject, aTopic, aData) {
Services.obs.removeObserver(this, "remote-settings:changes-poll-start");
const { expectedTimestamp } = JSON.parse(aData);
console.log("Polling started", expectedTimestamp ? "from push broadcast" : "by scheduled trigger");
},
};
Services.obs.addObserver(observer, "remote-settings:changes-poll-start");
Advanced Options
localFields
: records fields that remain local
During synchronization, the local database is compared with the server data. Any difference will be overwritten by the remote version.
In some use-cases it’s necessary to store some state using extra attributes on records. The localFields
options allows to specify which records field names should be preserved on records during synchronization.
const client = RemoteSettings("a-collection", {
localFields: [ "userNotified", "userResponse" ],
});
filterFunc
: custom filtering function
By default, the entries returned by .get()
are filtered based on the JEXL expression result from the filter_expression
field. The filterFunc
option allows to execute a custom filter (async) function, that should return the record (modified or not) if kept or a falsy value if filtered out.
const client = RemoteSettings("a-collection", {
filterFunc: (record, environment) => {
const { enabled, ...entry } = record;
return enabled ? entry : null;
}
});
Debugging and manual testing
Logging
In order to enable verbose logging, set the log level preference to debug
.
Services.prefs.setStringPref("services.settings.loglevel", "debug");
Remote Settings Dev Tools
The Remote Settings Dev Tools extension provides some tooling to inspect synchronization statuses, to change the remote server or to switch to preview mode in order to sign-off pending changes. More information on the dedicated repository.
Preview Mode
Enable the preview mode in order to preview changes to be reviewed on the server. This can be achieved using the Remote Settings Dev Tools, or programmatically with:
RemoteSettings.enablePreviewMode(true);
In order to pull preview data on startup, or in order to persist it across restarts, set services.settings.preview_enabled
to true
in the profile preferences (ie. user.js
).
For release and ESR, for security reasons, you would have to run the application with the MOZ_REMOTE_SETTINGS_DEVTOOLS=1
environment variable for the preference to be taken into account. Note that toggling the preference won’t have any effect until restart.
Trigger a synchronization manually
The synchronization of every known remote settings clients can be triggered manually with pollChanges()
:
await RemoteSettings.pollChanges()
In order to ignore last synchronization status during polling for changes, set the full
option:
await RemoteSettings.pollChanges({ full: true })
The synchronization of a single client can be forced with the .sync()
method:
await RemoteSettings("a-key").sync();
Important
The above methods are only relevant during development or debugging and should never be called in production code.
Inspect local data
The internal IndexedDB of Remote Settings can be accessed via the Storage Inspector in the browser toolbox.
For example, the local data of the "key"
collection can be accessed in the remote-settings
database at Browser Toolbox > Storage > IndexedDB > chrome, in the records
store.
Delete all local data
All local data, of every collection, including downloaded attachments, can be deleted with:
await RemoteSettings.clearAll();
Unit Tests
As a foreword, we would like to underline the fact that your tests should not test Remote Settings itself. Your tests should assume Remote Settings works, and should only run assertions on the integration part. For example, if you see yourself mocking the server responses, your tests may go over their responsibility.
If your code relies on the "sync"
event, you are likely to be interested in faking this event and make sure your code runs as expected. If it relies on .get()
, you will probably want to insert some fake local data.
Simulate "sync"
events
You can forge a payload
that contains the events attributes as described above, and emit it :)
const payload = {
current: [{ id: "abc", age: 43 }],
created: [],
updated: [{ old: { id: "abc", age: 42 }, new: { id: "abc", age: 43 }}],
deleted: [],
};
await RemoteSettings("a-key").emit("sync", { data: payload });
Manipulate local data
A handle on the underlying database can be obtained through the .db
attribute.
const db = RemoteSettings("a-key").db;
And records can be created manually (as if they were synchronized from the server):
const record = await db.create({
id: "a-custom-string-or-uuid",
domain: "website.com",
usernameSelector: "#login-account",
passwordSelector: "#pass-signin",
});
If no timestamp is set, any call to .get()
will trigger the load of initial data (JSON dump) if any, or a synchronization will be triggered. To avoid that, store a fake timestamp. We use Date.now()
instead of an arbitrary number, to make sure it’s higher than the dump’s, and thus prevent its load from the test.
await db.importChanges({}, Date.now());
In order to bypass the potential target filtering of RemoteSettings("key").get()
, the low-level listing of records can be obtained with collection.list()
:
const { data: subset } = await db.list({
filters: {
"property": "value"
}
});
The local data can be flushed with clear()
:
await db.clear()
Misc
We host more documentation on https://remote-settings.readthedocs.io/, on how to run a server locally, manage attachments, or use the REST API etc.
About blocklists
The security settings, as well as addons, plugins, and GFX blocklists were the first use-cases of remote settings, and thus have some specificities.
For example, they leverage advanced customization options (bucket, content-signature certificate, target filtering etc.). In order to get a reference to these clients, their initialization code must be executed first.
const {RemoteSecuritySettings} =
ChromeUtils.importESModule("resource://gre/modules/psm/RemoteSecuritySettings.sys.mjs");
RemoteSecuritySettings.init();
const {BlocklistPrivate} =
ChromeUtils.importESModule("resource://gre/modules/Blocklist.sys.mjs");
BlocklistPrivate.ExtensionBlocklistRS._ensureInitialized();
BlocklistPrivate.PluginBlocklistRS._ensureInitialized();
BlocklistPrivate.GfxBlocklistRS._ensureInitialized();
Then, in order to access a specific client instance, the bucketName
must be specified:
const client = RemoteSettings("onecrl", { bucketName: "security-state" });
And in the storage inspector, the IndexedDB internal store will be prefixed with security-state
instead of main
(eg. security-state/onecrl
).